Energy storage equipment anomaly detection method based on KANs algorithm

Abstract

The invention discloses an energy storage equipment anomaly detection method based on a KANs algorithm, and relates to the technical field of energy storage equipment anomaly detection, and the method comprises the steps: obtaining multi-dimensional time series data affecting the operation of energy storage equipment in real time; preprocessing the multi-dimensional time sequence data; a KANs anomaly detection model is constructed on the basis of the Kolmogorov-Arnold theorem; on the basis of the preprocessed data and a KANs anomaly detection model, abnormal behaviors of the energy storage equipment are judged; if the abnormal behavior exists, the abnormal behavior of the energy storage equipment is solved; if no abnormal behavior exists, the energy storage equipment operates normally; according to the method, the detection accuracy can be improved, the false alarm rate can be reduced, and damage to energy storage equipment and safety accidents can be avoided or reduced by predicting potential faults and taking preventive measures in advance.

Description

Technical Field

[0001] The present invention relates to the technical field of abnormality detection of energy storage equipment, and in particular to an abnormality detection method of energy storage equipment based on a KANs algorithm. Background Art

[0002] In modern power systems, the role of energy storage equipment is becoming increasingly significant. With the widespread access to renewable energy and the increase in grid load, energy storage equipment is not only responsible for basic functions such as peak load regulation, frequency regulation and backup power supply, but also plays a key role in the stable operation and optimized dispatch of power systems.

[0003] However, energy storage devices face a variety of potential abnormal risks under extreme operating conditions (such as high load, high temperature, etc.), including battery capacity attenuation, thermal runaway, electrical failure, etc. Traditional abnormality detection methods usually rely on thresholds and simple rule-based detection systems. Although simple to operate, they often lack flexibility and accuracy when dealing with complex working conditions and hidden abnormalities, which may lead to serious safety accidents and economic losses.

[0004] The Chinese patent (CN118362923A) discloses a method and system for energy storage safety warning based on multi-modal perception. Although this method solves the problems of single-modal data processing, imperfect abnormal warning mechanism, and lack of multi-level abnormal identification and safety warning in the prior art, it provides a more reliable guarantee for the safe operation of energy storage battery packs.

[0005] However, when dealing with complex multidimensional data, the generalization ability of the model is still insufficient and it is easily affected by data noise and dimensionality disaster, resulting in high false alarm and missed alarm rates. In addition, this method mainly relies on predefined features and rules, and has limited detection effects on hidden faults and early anomalies, and cannot flexibly cope with the complex operating environment of new energy storage equipment. These problems are particularly evident when dealing with anomaly detection of large-scale energy storage equipment;

[0006] Therefore, we propose a detection method that can improve the accuracy of detection, reduce the false alarm rate, and take preventive measures in advance by predicting potential failures to avoid or mitigate damage to energy storage equipment and safety accidents. Summary of the invention

[0007] In order to overcome the deficiencies in the background technology, the present invention discloses a method for detecting abnormality of energy storage equipment based on a KANs algorithm.

[0008] In order to achieve the above-mentioned invention object, the present invention adopts the following technical scheme:

[0009] A method for detecting abnormality of energy storage equipment based on KANs algorithm, the specific steps are as follows:

[0010] Real-time acquisition of multi-dimensional time series data that affects the operation of energy storage equipment;

[0011] Preprocessing multidimensional time series data;

[0012] Construct a KANs anomaly detection model based on the Kolmogorov-Arnold theorem;

[0013] Based on the preprocessed data and KANs anomaly detection model, the abnormal behavior of the energy storage device is judged;

[0014] If there is abnormal behavior, solve the abnormal behavior of the energy storage device;

[0015] If there is no abnormal behavior, the energy storage device operates normally.

[0016] Preferably, the multi-dimensional time series data includes voltage, current, temperature and charge and discharge state parameters.

[0017] Preferably, after acquiring the multidimensional time series data, the integrity and consistency of the detection data are checked. When the data is incomplete, an interpolation algorithm is used to compensate the acquired or time series data, and the calculation formula of the interpolation algorithm is:

[0018]

[0019] in, Represents multidimensional time series data x i (t) Compensated data at time t.

[0020] Preferably, the preprocessing step includes denoising, outlier processing and normalization processing.

[0021] Preferably, a low-pass filter is used to perform the denoising process.

[0022] Preferably, the outliers are detected and processed using sliding window technology and statistical technology, wherein the calculation formula is:

[0023] Assuming the size of the sliding window is W, for each window data, calculate the multidimensional time series data x i The mean value μ of (t) i and standard deviation σ i ;

[0024]

[0025] The normalization process adopts Z-score normalization, and its calculation formula is:

[0026]

[0027] Preferably, the KANs anomaly detection model is used to decompose multidimensional time series data into a combination of multiple univariate functions, and the calculation formula is as follows:

[0028]

[0029] Among them, φ i and ψ ij is a one-dimensional continuous function, X is the input multidimensional time series data, and X=(x1,x2,x3…,x n ).

[0030] Preferably, the step of determining abnormal behavior of the energy storage device is specifically as follows:

[0031] 1) Substitute the preprocessed data into the KANs anomaly detection model for calculation;

[0032] 2) Comparing the calculation result with the dynamic threshold;

[0033] 3) If the comparison result exceeds the threshold, it is determined that the energy storage device has abnormal behavior, and the abnormal behavior of the energy storage device is resolved;

[0034] 4) The comparison result is that the threshold is not exceeded, and it is determined that the energy storage device has no abnormal behavior and the energy storage device will operate normally.

[0035] Preferably, the method for solving abnormal behavior of the device is:

[0036] 1) Based on the preprocessed data, determine the degree of abnormal behavior of the energy storage device;

[0037] 2) Establish a hierarchical protection mechanism and implement corresponding solutions based on the different degrees of abnormal behavior.

[0038] Preferably, the specific calculation formula for determining the degree of abnormal behavior of the energy storage device is:

[0039]

[0040] Due to the adoption of the above-mentioned technical solution, the present invention has the following beneficial effects:

[0041] The present invention discloses a method for detecting anomalies of energy storage devices based on a KANs algorithm, which can effectively improve the data quality by preprocessing the time series data, so that when anomaly detection models are subsequently used to detect anomalies of energy storage devices, more accurate detection results can be obtained and the false alarm rate can be reduced.

[0042] In addition, by real-time detection of the operating status of the energy storage equipment and timely resolution of abnormal behaviors of the energy storage equipment, the probability of damage to the energy storage equipment and safety accidents can be effectively avoided or reduced. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 A schematic diagram of a method of the present invention;

[0044] Figure 2 A schematic diagram of the workflow of this method. DETAILED DESCRIPTION

[0045] The technical solution of the present invention will be described below in conjunction with the drawings in the embodiments of the present invention. In the description, it should be understood that if there are terms such as "upper", "lower", "front", "back", "left", "right" and the like indicating directions or positional relationships, they are only corresponding to the drawings of the present invention, for the convenience of describing the present invention, and do not indicate or imply that the devices or components referred to must have a specific direction:

[0046] Embodiment 1 is:

[0047] Combined with Figure 1-2 The energy storage device abnormality detection method described above based on the KANs algorithm comprises the following specific steps:

[0048] Real-time acquisition of multi-dimensional time series data that affects the operation of energy storage equipment;

[0049] The multi-dimensional time series data is generally acquired in real time through a multi-type sensor network deployed at key locations of energy storage equipment. In particular, the multi-dimensional time series data includes voltage, current, temperature, and charge and discharge state parameters.

[0050] In addition, to ensure data consistency, a time synchronization mechanism, such as the Network Time Protocol (NTP) or the Global Positioning System (GPS), is introduced when acquiring data to unify the data collection timestamps of each sensor;

[0051] The calculation formula for data acquisition is:

[0052] Assume that the acquisition time of each sensor is t i , then through the time synchronization mechanism, the acquisition time of each sensor will be adjusted during data acquisition to meet the following conditions:

[0053] t i =t0+δ i ;

[0054] Among them, t0 is the standard time base, δ i is the time deviation of each sensor, which is calculated by NTP or GPS. i Perform calibration to ensure that the data collected by all sensors are accurately aligned.

[0055] Moreover, the data stream of each sensor is transmitted in an independent channel, which can be expressed as:

[0056] D i (t) = {d i1 (t),d i2 (t),…,d in (t)};

[0057] Among them, d i1 (t) represents the data stream of the i-th sensor at time t, d ij (t) is the data collected by the jth sensor channel at time t,

[0058] After the sensor network completes the collection, it continuously monitors the integrity and consistency of the data stream to achieve the purpose of real-time monitoring of the collection status. If an abnormality is detected in a data stream, such as data loss or communication interruption, the abnormal time point is automatically recorded and the data recovery mechanism is triggered;

[0059] As for the case of data loss, assuming that data loss occurs at time t, an interpolation algorithm is used to compensate for the data. The expression of the interpolation algorithm is:

[0060]

[0061] in, Represents multidimensional time series data x i (t) Compensated data at time t.

[0062] Preprocessing multidimensional time series data: After the data collection of the energy storage device is completed, the collected raw data usually contains noise and outliers, which may be caused by sensor failure, environmental interference or communication errors. Therefore, in order to ensure the accuracy and consistency of the data, the data is preprocessed to ensure the quality of the data.

[0063] Construct a KANs anomaly detection model based on the Kolmogorov-Arnold theorem;

[0064] The model is used to perform function decomposition on multidimensional data and extract deep-level features. By learning historical data, it can accurately identify the potential abnormal patterns of the equipment. After preprocessing, the data is generated into standardized data, which is input into the processing module of the KANs anomaly detection model. The KANs anomaly detection model decomposes multidimensional data into a combination of multiple single-variable functions and extracts deep nonlinear features in the data, thereby accurately identifying the potential abnormal behavior of energy storage equipment.

[0065] Based on the preprocessed data and KANs anomaly detection model, the abnormal behavior of the energy storage device is judged;

[0066] If there is abnormal behavior, solve the abnormal behavior of the energy storage device;

[0067] If there is no abnormal behavior, the energy storage device operates normally.

[0068] Among them, the abnormal behavior of the energy storage equipment is judged through a multi-level anomaly detection mechanism, and the intelligent alarm system can sound an alarm when abnormal behavior occurs, thereby ensuring that abnormal situations can be identified quickly and accurately, and certain solutions can be taken.

[0069] Embodiment 2 is:

[0070] Based on Example 1, the preprocessing step is further limited, that is, the preprocessing step includes denoising, outlier processing and normalization processing.

[0071] In particular, the denoising process is performed using a low-pass filter;

[0072] The transfer function of the low-pass filter is expressed as:

[0073]

[0074] H(f) is the filtered signal, f is the signal frequency, f c is the cut-off frequency, and n is the order of the filter. By adjusting the cut-off frequency f c , you can control the strength of the filter.

[0075] In addition, in data collection, some outliers occasionally appear, which may affect the accuracy of subsequent analysis and models. Therefore, sliding window technology and statistical technology are used to detect and process outliers. The calculation formula is:

[0076] Assuming the size of the sliding window is W, for each window data, calculate the multidimensional time series data x i The mean value μ of (t) i and standard deviation σ i ;

[0077]

[0078] The normalization process adopts Z-score normalization, and its calculation formula is:

[0079]

[0080] And after completing denoising, outlier processing and normalization, the preprocessing system will convert the format of the data to meet the input requirements of the KANs anomaly detection model, and the data is converted into a standard time series format to ensure that no important information is lost during the model processing.

[0081] Embodiment 3 is:

[0082] On the basis of Example 2, the KANs anomaly detection model is further limited, that is, the KANs anomaly detection model is used to decompose multidimensional time series data into a combination of multiple single variable functions, and the calculation formula is as follows:

[0083]

[0084] Among them, φ i and ψ ij is a one-dimensional continuous function, X is the input multidimensional time series data, and X=(x1,x2,x3…,x n ); This formula is based on the Kolmogorov-Arnold theorem, which shows how to decompose a multidimensional function f(X) into a combination of several single-variable functions, thereby simplifying the complexity of high-dimensional data. The formula is summed up by the outer layer and inner sum Double decomposition, sequentially multi-dimensional input X=(x1,x2,x3…,x n ) is converted into multiple univariate continuous functions φ i and ψ ij combination.

[0085] In this process, although the formula simplifies high-dimensional data into a combination of multiple single-variable functions, it can still capture the deep nonlinear characteristics in the data, thereby ensuring the accuracy and robustness of the model when processing complex data. This decomposition method helps to identify potential anomalies in energy storage devices, especially for anomaly detection of high-dimensional, nonlinear data. It not only improves the accuracy of detection, but also effectively reduces the false alarm rate and missed alarm rate, thereby providing a more reliable guarantee for fault detection of energy storage devices.

[0086] Embodiment 4 is:

[0087] On the basis of Example 1, the step of determining the abnormal behavior of the energy storage device is further limited, that is, the step of determining the abnormal behavior of the energy storage device is specifically:

[0088] 1) Substitute the preprocessed data into the KANs anomaly detection model for calculation;

[0089] 2) Comparing the calculation result with the dynamic threshold;

[0090] 3) If the comparison result exceeds the threshold, it is determined that the energy storage device has abnormal behavior, and the abnormal behavior of the energy storage device is resolved;

[0091] 4) The comparison result is that the threshold is not exceeded, and it is determined that the energy storage device has no abnormal behavior and the energy storage device will operate normally.

[0092] Here, a dynamic threshold mechanism is adopted. The dynamic threshold can be adjusted according to the operating status of the device, the distribution of historical data, and the change trend of real-time data, so as to better adapt to the changing operating environment; the setting of the dynamic threshold T(t) can be based on the following formula:

[0093] T(t)=μ f +kσ f

[0094] Among them, μ f The historical mean of the output features of the KANs anomaly detection model. f is the standard deviation of the output feature. k is the adjustment coefficient, which is used to control the sensitivity of the threshold. When making judgments, the value of k can be dynamically adjusted according to the operating conditions and historical performance of the energy storage device to achieve the best detection effect. Through dynamic adjustment, false positives and false negatives can be reduced, especially when the operating status of the energy storage device changes, the dynamic threshold can respond to changes in a timely manner and provide more accurate anomaly detection.

[0095] Embodiment 5 is:

[0096] On the basis of Example 1, the method for solving the abnormal behavior of the device is further limited, that is, the method for solving the abnormal behavior of the device is:

[0097] 1) Based on the preprocessed data, determine the degree of abnormal behavior of the energy storage device;

[0098] 2) Establish a hierarchical protection mechanism and implement corresponding solutions based on the different degrees of abnormal behavior.

[0099] Among them, this step is only started when the KANs anomaly detection model determines that the energy storage device is in an abnormal operating state, so as to start more intensive monitoring of the operating state of the energy storage device. This module continuously collects and analyzes real-time data, focusing on key parameters that may cause further damage to the equipment (such as temperature, current, voltage, etc.), and compares these data with historical data to determine the development trend of the anomaly and the current state of the equipment;

[0100] For real-time monitoring data x i (t), calculate its rate of change Δx i (t) and the deviation θ relative to historical data are used to determine the degree of abnormal behavior of the energy storage device. The specific calculation formula is:

[0101]

[0102] Among them, μ i and σ iThey are the mean and standard deviation of historical data, respectively. These calculations help determine whether the current anomaly is worsening and whether further emergency measures are needed;

[0103] After that, a hierarchical protection mechanism is established. Depending on the severity of the anomaly detection, different levels of protection mechanisms are activated to prevent further damage to the device. They are divided into the following categories:

[0104] 1) Level 1 protection: Applicable to situations where there are minor anomalies or the operating status of the energy storage device slightly deviates from the normal range. At this time, a warning signal may be issued and the device operating parameters will be recorded in the background, but the operating status of the device will not be intervened immediately.

[0105] 2) Secondary protection: Applicable to situations where the abnormality is more serious but has not yet threatened the safety of the equipment. By performing power reduction operations, the load of the equipment is reduced, and the output of current or voltage is reduced to prevent the equipment from being damaged due to continuous high-load operation.

[0106] 3) Level 3 protection: Applicable to situations where there are serious anomalies or the equipment is in a high-risk state. At this time, the equipment will be immediately switched to safe mode or backup mode, and emergency shutdown operations may be triggered to prevent serious equipment failures or safety accidents.

[0107] 4) All-round protection: In extreme cases, if an abnormality is detected that may cause equipment damage or create a security threat (such as explosion, fire, etc.), it is necessary to immediately disconnect all key connections and cut off the power supply to the equipment to ensure the safety of the equipment and the surrounding environment.

[0108] In addition, the method will continuously optimize the anomaly detection and protection strategies based on historical records and current detection results. For example, the threshold can be adjusted based on the anomaly pattern that occurs repeatedly, or the triggering conditions of the protection measures can be optimized based on the operation history of the device. The optimized strategy will make the method more accurate and efficient in future anomaly detection.

[0109] The parts of the present invention that are not described in detail are prior art. It is obvious to those skilled in the art that the present invention is not limited to the details of the above-mentioned exemplary embodiments, and the present invention can be implemented in other specific forms without departing from the spirit or basic characteristics of the present invention; therefore, no matter from which point of view, the embodiments should be regarded as exemplary and non-restrictive. The scope of the present invention is limited by the appended claims rather than the above description. Therefore, it is intended to include all changes within the meaning and scope of the equivalent elements of the claims in the present invention, and any figure marks in the claims should not be regarded as limiting the claims involved.

Claims

1. A method for detecting abnormality of energy storage equipment based on KANs algorithm, characterized by: Specific steps for: Real-time acquisition of multi-dimensional time series data that affects the operation of energy storage equipment; Preprocessing multidimensional time series data; Construct a KANs anomaly detection model based on the Kolmogorov-Arnold theorem; Based on the preprocessed data and KANs anomaly detection model, the abnormal behavior of the energy storage device is judged; If there is abnormal behavior, solve the abnormal behavior of the energy storage device; If there is no abnormal behavior, the energy storage device operates normally.

2. The method for detecting abnormality of energy storage equipment based on the KANs algorithm according to claim 1, characterized in that: The multi-dimensional time series data includes voltage, current, temperature and charge and discharge state parameters.

3. The method for detecting abnormality of energy storage equipment based on the KANs algorithm as claimed in claim 2, characterized in that: After acquiring the multidimensional time series data, the integrity and consistency of the detection data are tested. When the data is incomplete, an interpolation algorithm is used to compensate the acquired or time series data, and the calculation formula of the interpolation algorithm is: in, Represents multidimensional time series data x i (t) Compensated data at time t.

4. The method for detecting abnormality of energy storage equipment based on the KANs algorithm according to claim 1, characterized in that: The preprocessing steps include denoising, outlier processing and normalization processing.

5. The method for detecting abnormality of energy storage equipment based on the KANs algorithm as claimed in claim 4, characterized in that: The denoising process is performed using a low-pass filter.

6. The method for detecting abnormality of energy storage equipment based on the KANs algorithm as claimed in claim 4, characterized in that: The outliers are detected and processed using sliding window technology and statistical technology, where the calculation formula is: Assuming the size of the sliding window is W, for each window data, calculate the multidimensional time series data x i The mean μ of (t) i and standard deviation σ i ; The normalization process adopts Z-score normalization, and its calculation formula is:

7. The method for detecting abnormality of energy storage equipment based on the KANs algorithm according to claim 1, characterized in that: The KANs anomaly detection model is used to decompose multidimensional time series data into a combination of multiple univariate functions. The calculation formula is as follows: Among them, φ i and ψ ij is a one-dimensional continuous function, X is the input multidimensional time series data, and X=(x1,x2,x3…,x n ).

8. The method for detecting abnormality of energy storage equipment based on the KANs algorithm according to claim 1, characterized in that: The steps for determining abnormal behavior of the energy storage device are specifically as follows: 1) Substitute the preprocessed data into the KANs anomaly detection model for calculation; 2) Comparing the calculation result with the dynamic threshold; 3) If the comparison result exceeds the threshold, it is determined that the energy storage device has abnormal behavior, and the abnormal behavior of the energy storage device is resolved; 4) The comparison result is that the threshold is not exceeded, and it is determined that the energy storage device has no abnormal behavior and the energy storage device will operate normally.

9. The method for detecting abnormality of energy storage equipment based on the KANs algorithm according to claim 1, characterized in that: The method for resolving abnormal device behavior is: 1) Based on the preprocessed data, determine the degree of abnormal behavior of the energy storage device; 2) Establish a hierarchical protection mechanism and implement corresponding solutions based on the different degrees of abnormal behavior.

10. The method for detecting abnormality of energy storage equipment based on the KANs algorithm according to claim 9, characterized in that: The specific calculation formula for determining the degree of abnormal behavior of the energy storage device is:

Images